Forecasting solar cycle 24 using neural networks

This paper presents the development of a model to predict solar cycle (SC) 24 using the technique of neural networks (NNs). The model is provided with the time input parameters defining the year and the month of a particular SC, in order to characterise the temporal behaviour of sunspots as observed in the last 10 SCs. Geomagnetic activity measured by the aa index is also introduced as an input to the NN model in such a way that the model establishes the relationship between the aa index in the previous SC with the sunspot number (SSN) of the following SC. Assuming January 2008 as the minimum for SC 24, the shape and amplitude of SC 24 is estimated in terms of monthly mean and smoothed monthly SSN. This new prediction model estimates an average SC 24, with the maximum occurring around June 2012 [±10 months] with a smoothed monthly maximum SSN of 117.5±8.5.